Year: 2026
Type: preprint
Abstract: Previously, we developed a statistical framework and software (PolyOrigin1) to reconstruct parental haplotypes of each offspring in connected tetraploid F1 populations. A key application of the technology is quantitative trait loci (QTL) mapping, via regression of phenotypes on posterior probabilities of parental origins. In this work, we extend the methodology and software (PolyOrigin2) to multi-generational populations (both diploid and tetraploid), which are common in breeding programs. The extension links parental origins across generations to estimate posterior probabilities of founder origins. Compared to PolyOrigin1, the main challenge is to phase intermediate individuals that are both parents and offspring, which is overcome by the Bayesian update rule. In addition, genotypic error models are extended to account for sequencing errors such as allelic bias and overdispersion in genotyping-by-sequencing (GBS). Extensive simulations were used to compare PolyOrigin2 with previous methods and to study the effects of such variables as sequencing errors and the number of progeny per parent. We further evaluated PolyOrigin2 using three real datasets: diploid potato with 23 founders and 7 generations, tetraploid potato with 10 founders and 3 generations, and tetraploid blueberry with 10 founders and 4 generations. The results show that PolyOrigin2 is accurate for haplotype phasing and reconstruction in diploid and tetraploid pedigrees and opens new opportunities for haplotype-based QTL mapping.
Language: English
Subfield: Genetics
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7236726864
